Imprint Defect Detection Using Inference Model
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The existing imprint techniques face challenges in efficiently detecting and distinguishing between extrusion and unfilling defects during the pattern formation process on substrates, which are critical for producing high-quality microstructured devices.
Innovation Solution
An evaluation apparatus that processes images of the formed composition on a substrate to detect abnormalities using an inference model, determining the type and location of defects, and automatically classifying images as normal or abnormal.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual inspection of observation images is performed to detect extrusion and unfilling defects, then detection accuracy can be maintained, but inspection time and labor cost increase significantly
Solution Approach 1:
The patent replaces manual mechanical inspection with an automated image processing system that uses computer algorithms to detect and classify defects. The system automatically analyzes observation images to identify extrusion and unfilling defects, eliminating the need for manual inspection while maintaining detection accuracy and significantly reducing inspection time.
Solution Approach 2:
The patent introduces an image processing system as an intermediary between the imprint process and quality assessment. This intermediary system captures observation images, processes them through automated algorithms, and provides defect detection results, serving as a bridge that eliminates direct manual inspection while preserving quality control.
2Measurement precision
If high-powered microscope is used to obtain detailed observation images for defect detection, then measurement precision improves, but the observation range becomes narrow requiring more images to be checked
Solution Approach 1:
The patent segments the inspection task into multiple observation images that collectively cover the entire shot region. By dividing the large-area inspection into smaller, manageable image segments and processing them automatically, the system achieves both high detection precision and complete area coverage without requiring manual inspection of numerous images.
Solution Approach 2:
The patent transitions from manual two-dimensional image inspection to automated multi-dimensional processing by combining multiple observation images into a comprehensive analysis. The system processes images in a systematic manner across different regions and depths, effectively expanding the observation range while maintaining precision through automated algorithms.
3Reliability
If multiple observation images are captured to cover the entire shot region, then complete defect detection is achieved, but the complexity of processing and analyzing the images increases
Solution Approach 1:
The patent employs a universal image processing system that handles multiple functions including image capture, processing, defect detection, and classification across all observation images. This multi-functional system simplifies the overall complexity by providing a unified approach to handle the entire inspection workflow rather than separate specialized processes for each task.
Solution Approach 2:
The image processing system performs self-service by automatically analyzing observation images without requiring manual intervention. The automated algorithms independently detect defects, classify them as extrusion or unfilling, and generate inspection results, thereby reducing processing complexity and eliminating the need for complex manual analysis procedures.
4Manufacturing precision
If various types of formation defects are detected and classified, then quality control improves, but the complexity of distinguishing and processing different defect types increases
Solution Approach 1:
The patent applies local quality analysis by detecting and classifying defects based on their specific characteristics and locations within the pattern. The system identifies different defect types (extrusion, unfilling) and their positions, providing localized quality assessment that improves manufacturing precision without requiring complex global analysis of all defects simultaneously.
Solution Approach 2:
The patent uses parameter changes in the image processing algorithms to automatically distinguish between different defect types. By varying analysis parameters such as shape characteristics, size thresholds, and positional relationships, the system efficiently classifies defects without requiring complex manual differentiation procedures, thereby improving quality control while managing processing complexity.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This solution enables efficient detection and classification of extrusion and unfilling defects, reducing manual effort and improving the quality of pattern formation by automating the evaluation process.
Implementation Method 1
a processing device configured to process the obtained image for the evaluation
Data Source
AI summary
An evaluation apparatus that evaluates a composition formed on a substrate by forming processing is provided. The apparatus comprises an obtaining device that obtains an image including the composition by the forming processing, and a processing device that processes the image for the evaluation. The processing device outputs a feature of each of one or more abnormalities in the image according to an inference model, obtains information regarding a formation region on the substrate where the composition has been formed, determines the kind of each of the abnormalities based on the output feature of each of the abnormalities and a relationship between the information and a position and a size of the abnormality, and makes, based on a result of the determination, final determination as to whether the image is a normal image or an image including an abnormality.